New Data Clustering Algorithm Combined of Ant Colony Algorithm and Improved Fuzzy C-Means Algorithm
Zhiming Zhang, Guobin Wu, Jie Luo · 2016
A new clustering algorithm combined of ant colony and improved Fuzzy C-Means (AC-LFCM) was proposed to resolve the shortage of Fuzzy C-Means (FCM) clustering algorithm on the presence of sensitive to initialization, easy to fall into local optimum and neglected the influence of local information of data.Firstly, aimed at the existing defects of FCM, a new algorithm named local-Fuzzy C-Means (LFCM) was formed through considering influence of data's neighborhood to target function; then introduced ant colony algorithm with great ability for disposing local extremum and parallel computing to fix on the initial numbers of clustering as well as the centers of clustering, combined with LFCM algorithm to find the whole distributing optimization clustering and achieve clustering analysis.And In the data clustering experiments on synthetic datasets and three datasets of UCI datasets by the LFCM and AC-LFCM algorithm, the results show that, compared with FCM, the algorithm has obvious advantage on the clustering performance.